hebrew and north west semitic: … and north west semitic: reflections on the classification of the...

24
HEBREW AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1. AS IS WELL KNOWN, the comparative method has been elaborated upon with reference to the lndo-European languages. For more than a century, it has been customary to view them from the angle of both the family-tree theory and the wave hypothesis. As far as the continuity of the territory of Indo- European languages can be posited, it is the wave hypothesis that best ex- plains the relation of the languages involved. Yet the position of the Indo- European languages in historical times presupposes migrations, and the lin- guistic situation due to them is best interpreted in the light of the family-tree theory . 1 2. One immediately recognizes that the wave hypothesis is much more con- vincing if applied to the Semitic languages. Even a hasty glance at the map of the Semitic tongues, in the past and in our own day, more or less reveals the continuity of the domain of these languages. Moreover, close contact between various Semitic idioms is well attested throughout history, the more so since Semitic languages surprisingly often were established as linguae francae, used in preference to the spoken language. This was the case not only with the language of a great power, viz., Akkadian, in which, e.g., Canaanite princes wrote their diplomatic correspondence with Pharaoh around 1400 B.C.E. (the Tel el-Amama letters), but also with Phoenician, used in Karatepe and Zen- I. See, e.g., Porzig (1954, p. 28). 21

Upload: trancong

Post on 19-Apr-2018

222 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

HEBREW AND NORTH WEST SEMITIC:

REFLECTIONS ON THE CLASSIFICATION

OF THE SEMITIC LANGUAGES

By

JOSHUA BLAU

Hebrew University of Jerusalem

1. AS IS WELL KNOWN, the comparative method has been elaborated upon with reference to the lndo-European languages. For more than a century, it has been customary to view them from the angle of both the family-tree theory and the wave hypothesis. As far as the continuity of the territory of Indo­European languages can be posited, it is the wave hypothesis that best ex­plains the relation of the languages involved. Yet the position of the Indo­European languages in historical times presupposes migrations, and the lin­guistic situation due to them is best interpreted in the light of the family-tree theory .1

2. One immediately recognizes that the wave hypothesis is much more con­vincing if applied to the Semitic languages. Even a hasty glance at the map of the Semitic tongues, in the past and in our own day, more or less reveals the continuity of the domain of these languages. Moreover, close contact between various Semitic idioms is well attested throughout history, the more so since Semitic languages surprisingly often were established as linguae francae, used in preference to the spoken language. This was the case not only with the language of a great power, viz., Akkadian, in which, e.g., Canaanite princes wrote their diplomatic correspondence with Pharaoh around 1400 B.C.E. (the Tel el-Amama letters), but also with Phoenician, used in Karatepe and Zen-

I. See, e.g., Porzig (1954, p. 28).

21

Page 2: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

22 JOSHUA BLAU

jirli, and especially with Aramaic. The status of the latter as an international language is mirrored in the Hebrew Bible in the demand of Hezekiah's offi­cials from Rab-Shakeh to speak Aramaic, rather than Akkadian (2 Kgs 18:26). Cultural contact is well attested outside the domain of the lingua franca as well, one of its outstanding examples being the affinity of poetic structure in Hebrew and Ugaritic. Accordingly, linguistic phenomena were apt to spread to Semitic dialects which had different origins and histories, which thus eventually give the impression that they historically belonged to a separate branch of the Semitic languages, which had earlier seceded from the other branches of the Semitic speech community. In fact, however, the simi­larity of these dialects is due to contact and parallel development.

2.1 The possibility of parallel development is, it seems, a factor to be taken into consideration even more in Semitic than in Indo-European languages. The importance of the fundamental difficulty of distinguishing between initial identity and parallel development, which pertains to the very essence of comparative linguistics, was stressed for comparative Indo-European gram­mar by Meillet in a famous paper (l 958, pp. 36-43). Because of the very close affinity of Semitic tongues, which are not less similar to each other than languages belonging to one branch of Indo-European, constituting a very similar starting point for the various Semitic languages, one must allow for the possibility of parallel development on a larger scale than in Indo-European linguistics. 2

2. Accordingly, it is very difficult to distinguish between the diffusion of linguistic elements, in accordance with the wave hypothesis, and parallel (and even convergent) development. Thus, the features interpreted by Rabin in his important and stimulating paper ( 1963) as due to linguistic diffusion, are, to my mind, rather cases of parallel development, the assumption of linguistic diffusion being contravened by the fact that most of the features adduced by Rabin extend over very long periods. Constant attention to the possibility of parallel development may save the scholar from pitfalls such as Kutscher's suggestion (1971. pp. 389-390) that, if during the centuries of its coexistence with Canaanite-Hebrew, Aramaic has jettisoned those (consonantal) phonemes that were alien to Canaanite· Hebrew, one can think of only one reason-the fact that the Canaanite-Hebrew substratum was able to assert itself, imposing its phonemic set upon the Aramaic substratum. This assumption, however. though possible, is by no means necessary. The (synchronic) identity of the stock of consonantal phonemes of Official Aramaic with that of Canaanite-Hebrew may well be due to convergent development. Setting aside the special devel­opment of the laryngals/pharyngals, due to the Sumerian substratum, Akkadian has developed exactly the same stock of consonantal phonemes as Canaanite-Hebrew, and, with the partial exception of the shift s >s, even the individual sound shifts were identical with that of Canaanite-Hebrew, as against the different sound shifts in Aramaic! Therefore, the (synchroni­cally) identical stock of consonantal phonemes in Canaanite-Hebrew and Aramaic may well be due to the general drift, which exerted its influence on Akkadian as well.

Page 3: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

HEBREW AND NORTH WEST SEMITIC 23

2.2 The importance of contact and parallel development in Semitic compara­tive grammar does not, of course, entirely dispense with the need for the model of the family-tree diagram. The basic dvision of the Semitic languages into East Semitic ( = Akkadian) and West Semitic is best explained by the assumption of the early separation of Akkadian from the rest of the Semitic tongues, in accordance with the family-tree model. Later, to be sure, Akka­dian came into close contact with various Semitic languages, influencing them3 and being influenced by them, 4 so that even for Akkadian, the family­tree model does not suffice. 5

3. Introductions to comparative linguistics almost invariably mention the more favorable case, where the parent language is known from written rec­ords,6 viz., the Romance dialects. Therefore, it will be convenient to start with such a case among the Semitic languages, viz., the modem Arabic dialects. One will take care, to be sure, not to simply identify Classical Arabic 7 or even_ its pre-Islamic predecessor with the parent language of the Arabic dialects or to consider pre-Islamic standard Arabic and the parent language of the Arabic dialects as if they were "mother" and "daughter." It stands to reason that they were rather "sister" languages, closely related and mutually intelligible, so that it is not too difficult to reconstruct the proto­language of the Arabic dialects which, to be sure, was by no means homogeneous (cf. Fischer, 1959, p. vii). The same applies, however, to Proto-Romance as well (see Hall, 1972): Classical Latin differs in various respects from Proto-Romance and is decidedly not its direct ancestor, and one must take into account the dialectal division in Proto-Romance as well. And just as the comparative study of the Romance languages is especially impor­tant because of the light it sheds on the value of our inferences in the cases in which no record of the proto-language is available, so is the comparative study of the Arabic dialects.

3. See, e.g., Zimmern (1917), Kutscher (1971, pp. 386-387), and Kaufman (!974). 4. See, e.g .. for Aramaic influence: Rimalt (!932) and Kutscher (1971, pp. 356--358); for

"Amorite" influence: Moran (l 965, p. 62) and Kaufman (1974, pp. 22-27), who also mentions Aramaic influence; in general: Ungnad-Matous (1964, pp. 7-8).

5. Yet even a model like Southwonh's (1964). which takes into account independent isoglos­ses, does not suffice, since it loses sight of changes due to contact.

6. So, e.g., Bloomfield (1933. pp. 300ff). 7. I do not claim, of course, that Arabic has especially preserved the Proto-Semitic language

type. In my opinion (see already Bergstriisser, 1928, pp. l34ff), in the main, Arnbic mirrors the late Semitic language type. Similarly, nobody will claim that Latin. rather than any other lndo­European language, mirrors the ancient Inda-European language structure. Yet both Latin and Arabic have the great advantage of, more or less, serving as proto-Janguages.

Page 4: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

24 JOSHUA BLAU

3.1 Before starting with a (rather cursory) overall comparative view of the Arabic dialects (see Section 3.2 below), it will be expedient to adduce some features in which pre-Islamic standard Arabic and the parent language of the Arabic dialects (at least partly) differ. Whereas Classical Arabic uses · i ("yes"), both 'e (see, e.g., Barthelemy, 1935-69, s.v.) and 'aywa (<'ay wallahi; see, e.g., Dozy, 1881, s.v.,8 also borrowed into Turkish, which again influenced some Arabic dialects), occurring in some dialects, attest to * 'ay. As against Classical Arabic haka('Ja, many dialects suggest an etymon hiiki(jii (see Blanc, 1964, p. 199, note 161 ). Whereas the perfect forms of verba mediae geminatae terminating in suffixes beginning with a consonant are built like halalt(u} in Classical Arabic, in dialects hallayt and its devel­opments prev~l, already attested as ancient dialectal forms (see Kofler, 194 l, p. 65; Wright, 1896-1898, I, p. 69).

3.2 Relying on features like hallayt mentioned above, Ferguson (1959) po­sited a relatively homogeneous koine in the first centuries of the Muslim era, from which most modern sedentary dialects stem, especially those outside Arabia. Yet this theory has to be rejected, especially since the history of the Arabic dialects lacks a single linguistic center of prestige and communication (for particulars see Blau, l965a, pp. 12-17 and forthcoming, §§6.5, 6.6). Moreover, Bedouin dialects, even from Arabia, frequently evince features that are generally considered as peculiar to the sedentary vernaculars and in other cases fluctuate between what is presumably old Bedouin usage and a new one which corresponds to that of the sedentary dialects, thus revealing the imprint of the latter and/or convergent development. Thus the koine stands at the end of the linguistic development, rather than at its beginning, being due to the general drift and the diffusion of linguistic features. The main lesson to be learned from the growth of the modem Arabic dialects is the understanding of the extent of the diffusion of language forms, also favored by the basic similarity of the various dialects (which also facilitated parallel development). Not only phonetic features, like the shift of q to ·, spread over a vast dialect area, as well as lexemes, like hayk ("so"; see Blau, 1965a, p. 14), but morphological innovations as well, as ~allayt quoted above. It is generally assumed (see, e.g., Weinreich, 1953, pp. 31-37, 43-44 and Kaufman, 1974, p. 122), that, under favorable circumstances, even bound morphemes may be transferred from language to language; and in the case of the Arabic dialects (as well as in the case of close contact between Semitic languages), the

8. Fegha!i (1919, p. 15, note 5) offers a much less plausible etymology. This, however, does not affect the derivation of' ay from *' Q)', rather than from *' i.

Page 5: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

HEBREW AND NORTH WEST SEMITIC 25

circumstances were favorable indeed: because of the great similarity between these tongues the diffusion of even bound morphemes was made possible. Accordingly. even unmotivated morphological innovations9 shared by Semi­tic languages being in close contact are not necessarily valid for genetic classification, since they may be due to the diffusion resulting from close contact. This, of course, makes genetic classification that much more dif­ficult.

4. It seems that the basic division of the Semitic languages into East Ak­kadian) and West Semitic is sound. Akkadian is characterized in a positive way by the far-reaching influence of the Sumerian substratum, phonetically mirrored in the weakening of the laryngals/pharyngals, and in a negative way by the absence of the West Semitic perfect qatala and the internal passive. It could have happened, to be sure, that some Semitic dialect exhibiting the Akkadian type stative *qatilal*qatula (rather than the West Semitic type perfect qatala and passive qutila etc.) developed the West Semitic forms through close contact with a Western type dialect rather than by genetically belonging to the West Semitic type (cf. Hetzron, 1975, p. 108); and Hetzron (1976, p. I 05) even considers the possibility of independent innovation. Yet, in my opinion, the existence of qatala and qutila in all the Semitic lan­guages10 with the exception of Akkadian makes the assumption of a separate historical West Semitic entity for at least most of them absolutely imperative, in accordance with the theory accepted by most scholars.

4.1 An additional feature, historically a part of the structure of the West Semitic languages, yet perhaps originally absent from Akkadian, is the indi­cative "imperfect" yqtlu .11 Some scholars, to be sure, posited it for Proto-

9. For the importance of morphological features for genetic classification cf .• e.g .• Meillet (1958, pp. 91-92), Polotsky (1964, p. 360, note 10). and especially Hetzron (1974, pp. 181 194; 1975, pp. 107-108; 1976, pp. 89-108).

10. During the development of the West Semitic languages some, in tum, lost qatala and qutila. qatala. e.g., is absent from modem East Aramaic dialects, and in many Semitic languages the internal passive has been superseded by former reflexive verbal themes. In the latter case, however, the category of the passive is still present, although its formal expression has changed. Moreover, in both cases, these languages exhibit later developments, after they had historically possessed qatalalqurila. In Akkadian, however, the use of the stative mirrors a stage in which the perfect qara/a and the passive had not yet developed.

11. Again, it is absent from many West Semitic languages. This, however, was due to historical development as well, during which the imperfectyqtlu disappeared. We shall deal later (in Section 5) with the hypothesis that yqrlu formed a part of "Central'' Semitic only. Here, we are concerned only with the question of whether or not yqrlu was a part of the Proto-Akkadian verbal structure as well, i.e., whether it was a Proto-Semitic or rather a West Semitic feature.

Page 6: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

26 JOSHUA BLAU

Semitic (Soden, 1959, as punctual, as against durative yaqattil, both Proto­Semitic in his opinion), whereas Kurylowicz (1961, pp. 55ff; cf. also p. 49) considered yqtlu the original imperfect in Akkadian as well, which was later replaced in Akkadian by yaqatta/, etc. and confined to secondary functions, viz., as modus re/ativus (in relative clauses, etc.). Since the relegation of ancient features to secondary features (I would rather speak of their preserva­tion in closed syntagmemes)12 is very well attested, this proposition is quite attractive. One must not, however, lose sight of the fact that, synchronically at least, the alleged continuation of Proto-Semitic *yaqtulu, i.e., Akkadian iprusu, does not denote the indicative imperfect in relative clauses, etc., but rather the preterite in this syntactic environment, the imperfect (present) being marked by iparrasu. Accordingly, if one adheres to the view that iprusu originates in the Proto-Semitic indicative imperfectyaqtulu, one has to postu­late that, through the impact of the Akkadian preterite iprus, the use of iprusu became that of modus relativus of the preterite (and then the modus relativus ending -u was transposed to the other indicative forms as well). 13 This suppos­ition is, to be sure, possible, yet rather intricate. On the other hand, there are internal Akkadian indications which may suggest the (comparatively) late date of the emergence of Akkadian iparras ( =yaqattal): many "weak" verbs14

exhibit "weak" formation of iparras (see Soden, 1952, pp. 126ff) and this may indicate that these forms were built according to the preterite forms, exhibiting yaqtul etc. This assumption would, e.g., explain the present ubbal (from wabiilum, "to carry"), formed according to the preterite ubil (alongside "strong" D uwasfor; one will interpret "weak" D forms as being influenced by the "weak" G forms). This interpretation is, prima facie, buttressed by the fact that in Old Akkadian original I pharyngals/laryngals behave as "strong" verbs (see Gelb, 1961, p. 181 ), thus suggesting that their later "weakness" is due to the influence ofverba I w!y etc. Accordingly, one would fix the time of

12. For the relegation of ancient features to closed syntagmemes (cf. Blau, 1961-62, pp. 70-71) see Arabic yqtl in the sense of the past used only after lam and in conditional clauses, Hebrew yqtl after wa- ('"and"); Arabic Iii preceding qr/ in parallel negations; yktb (as against bktb) in some modern Arabic dialects as subjunctive; Arabic 'ahad (as against wiihid) after negation etc., as the first member of construct structures and in certain numerals; in modem Aramaic dialects the participle without vo is used in subordination (Cohen, 1924. p. 216). Cf. also Blau (1960, p. 100).

13. For the productivity of the -u suffix cf. its Middle Babylonian (sometimes also Old Babylonian) addition to the third person singular feminine ending of the stative (cf. Soden, 1952, p. 108). Sarauw (1912, p. 68) posits an original modus relativus iparrasu, from which the u was then transferred to iprus.

14. I do not take into account verba mediae wly, because the doubling of the last radical, rather than of wly. may be due to the aversion to double w, aitested in several Semitic languages (see Blau, 1971, pp. 147f0.

Page 7: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

HEBREW AND NORTH WEST SEMITIC 27

the emergence of iparras after the ''weak'' formation of G iprus of I wly (as demonstrated by ubbal built according to 1/bil), but before I pharyngals/ laryngals had become "weak" (as hinted by "strongly" formed iparras of these verbs in Old Akkadian). If the assumption of the late date of iparras should, in fact, tum out to be true, 15 it would make the notion of an early Akkadian *yaqtulu much more palatable.

4.2 There are, however, rather strong indications that yaqattal etc. is a Proto-Semitic, or even a "Hamito-Semitic" feature, being attested in Ethiopic and Berber (see Polotsky, 1964, pp. 110-111 ). Parallel develop­ment, to be sure, is a very frequent feature in Semitic languages (see Section 2.1 above) and it has been assumed for yaqattal by many scholars (see the literature adduced by Polotsky, 1964, p. 358, note 29). Yet one has to admit that the existence of a strange feature such as gemination outside D, 16 occur­ring in Akkadian, Ethiopic and Berber makes it much more likely to assume that Akkadian iparras (as well as its Ethiopic 17 and Berber parallels) continue a Proto-(Hamito-)Semitic feature, although it is not easy to understand its "weak" formation in the Akkadian weak verb.

Nevertheless, this assumption does not necessarily imply that the Akka­dian modus relativus iprusu cannot be the continuation of Proto-Semitic yaq­tulu etc. One could postulate (with Soden, 1957) that in Proto-Semitic two present/future forms coexisted, viz., punctual yaqtulu etc., and durative yaqattil etc. In Akkadian yaqattil superseded yaqtulu, which was preserved only in closed syntagmemes, i.e., in relative clauses etc.

5. In Ethiopic and modem South Arabic no traces of yaqtulu etc. can be detected, and we do not know how to vocalize the prefix tense forms in Epigraphic South Arabic. 18 The so-called North West Semitic languages as

15. Cf. also Bergstrasser, 1918-29, II, p. 12. 16. Cf. Sarauw (1912, p. 67). On the other hand. I do not agree with his assumption that.

were yaqtufu early, yaq1ulu andyaqtul would have to be interpreted as moods of one tense. The "imperfect" yaq111/11 and the preterite yaqruf are different tenses, and it is only between yaq111/11 and the jussive yaqtul that a relationship of moods obtains.

17. No certain traces of yaqattal etc. have been detected in North West Semitic languages. In "Amorite." because of its close contact with Akkadian. such forms, especially in proper names, could well exhibit Akkadian influence. Biblical Hebrew reflects no vestiges of it (despite Meyer, 1966-72, I. p. 19; II. pp. 121, 134-135), nor does Ugaritic (see Gordon, 1965, p. 67, §9.2, following H. L. Ginsberg). Cf., in general, Fenton (1970).

18. The subjunctive ofverba !If n'iy clearly evinces that it terminated in a vowelless w!y. As to the controversy on the origins of the Main Verb-Markers in the Northern Gurage verb, see the exposition and the literature adduced by Hetzron (1977, pp. 88-92). At any rate. even according to Hetzron's view, the Main Verb-Markers do not attest to Proto-West Semitic yaqtu/11.

Page 8: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

28 JOSHUA BLAU

well as Arabic, on the other hand, attest to yaqtulu. Therefore, Hetzron (1974, p. 189; 1975, p. 107; 1976, p. 105; 1977, pp. 14-15), who regardedyaqtulu as an innovation, posited a Central Semitic branch, consisting of Arabic, Canaanite and Aramaic, sharing the morphological innovation yaqtulu (cf. also Polotsky, 1964, p. 110), thus restricting South Semitic to South Arabic and Ethiopic, which, like Akkadian, did not share the innovated yaqtulu.

One can hardly consider the claim that yaqtulu was absent from Proto­South Arabic (from which Ethiopic as well branched off) as proven. If Proto­Semitic in fact had two forms marking the present/future, viz., punctual yaqtulu and durative yaqattil, these forms were threatened because of their similarity in function, on the one hand, and because of their formal similarity to yaqtul and the D prefix form, on the other. Therefore, the total absence of one of them from any language does not prove that it did not occur in a preceding stratum. Thus the total disappearance of the short imperfect from rabbinic Hebrew, as opposed to the ordinary imperfect, could be interpreted as an indication of the total absence of yaqtul u in the linguistic strata preced­ing rabbinic Hebrew; yet biblical Hebrew, exhibiting the short imperfect as opposed to the ordinary one, attests to the existence of yaqtulu. Similarly, nothing must be inferred from the lack of any indication of yaqtulu in modern South Arabic and Ethiopic for Proto-South Arabic. Furthermore, as to Epi­graphic South Arabic, although it is very difficult to state anything certain because of the total lack of vocalization, the possibility of w/y used as matres lectionis (see Beeston, 1962, p. 5; Jamme, 1962, p. 80b, line 7), and the uncertainty as to the classes of verbal themes, I nevertheless have the feeling that it is easier to interpret various forms by the assumption that, presumably alongside yaqattil etc., Epigraphic South Arabic also had yaqtulu etc. Thus the simplest way to account for verbs I I wly spelled without w(v in indicative function (like yknn; see Beeston, 1962, p. 26) is to postulate yakununna, yakimanna, i.e. an "energetic" form derived from yaqtul. Yet one may interpret such forms in accordance with Akkadian forms exhibiting doubling of the last, rather than of the middle radical (like ikunnu). The same may apply (cf. Jamme, 1962, no. 577, line 5) towyh'nw ("and they helped," root 'wn), which, however, may also exhibit short imperfect denoting the past (yet cf. ibid., line 12 ylfyhmw, also referring to the past, and because of they presumably not to be interpreted as short imperfect). On the other hand, yhqfJw (with assimilation of the n, root nq6, "he carried away"; Jamme, 1962, no. 586, line 22) by necessity exhibits a vowelless first radical (because of the assimilation of the n) and is, therefore, presumably to be interpreted as yaqtulu (this is more likely than its interpretation as short imperfect referring to the past, cf. yifyhmw quoted above). If the b-imperfect is in fact indicative

Page 9: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

HEBREW AND NORTH WEST SEMITIC 29

(see Hafner, 1943, pp. 79-81; Beeston, 1962, pp. 24-25), bymd (root mdd; Beeston, 1976, p. 420, line 72) exhibitsyaqtulu. Hafner (1943, p. 90) inter­prets yhzh as yuhazihhu, and I am inclined to interpret similarly the h-causatives of srr and s3' I referring to the past (see Jamme, 1962, s. vv. hsr and hsl [!]). I have also noted imperfect forms terminating in (energetic?) n in relative clauses in indicative function, without, however, losing sight of the fact that in Epigraphic South Arabic there seems to be formal affinity between the use of the imperfect in relative clauses and the jussive (see Beeston, 1962, p. 24): yknn (see above, root kwn) in Jamme (1962, no. 750, line 14); yz'n (root wz', "he will continue"), ibid., no. 577, line 18; ymrn (root mrr, "it occurs(?)"), ibid., no. 711, line 5. Cf. also Jamme (1962) no. 669, line 10; no. 729, line 9 yldn ("he will be born," in conditional clauses), root wld; no. 577, line 15 y8rw ("they leveled'', root w(Jr). These examples (which can easily be augm~nted), although they can be explained in various ways, nevertheless give the impression that they at least partly attest to the existence of yaqtulu in Epigraphic South Arabic. At any rate, there is no indication for its absence from Epigraphic South Arabic. Accordingly, one ought rather to abstain from transferring Arabic to the ''innovating Central Semitic'' branch because of the alleged absence of yaqtulu etc. in Proto-South Arabic.

5.1 Nor do the other alleged innovations of Arabic, shared with Canaanite and Aramaic, prove that Arabic is closer to Canaanite and Aramaic than to South Arabic and Ethiopic, and that, accordingly, it has to be grouped with the former as Central Semitic, rather than with the latter as South Semitic.

5.1.1 There exist positive indications of shared innovations common to South Arabic, Ethiopic, and Arabic. The most outstanding among these are the ''broken plurals.'' One must not claim that they mirror a common retention (so Hetzron, 1975, p. 102): even if the notion of broken plurals should tum out to be an ancient Hamito-Semitic feature, it is not the existence of broken plurals as such that proves the close affinity of Arabic, South Arabic and Ethiopic, but rather their widespread formal identity; cf., e.g., Arabic qital, Ge'ezq::ital; Arabic qatalat, Ge'ezqatalt, both denoting plurals of participles and nomina agentia; Arabic qitiil, Ge'ez q::itiil; Arabic qutul, Ge'ez q::itiil; Arabic 'aqtul, 'aqtilat, Ge'ez 'aqt::il, 'aqt::ilt; Arabic and Ge'ez 'aqtiil; Arabic maqatil, maqatilat, Ge'ez maqiit::il, maqatdlt. The occurrence of some scat­tered parallels to these broken plurals in other Semitic languages must not be considered remnants, but rather the primary elements from which the South Semitic languages, including Arabic, built their broken plurals. The compara­tively late age of the broken plurals is proven by their invariable tri-radical

Page 10: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

30 JOSHUA BLAU

form, whereas the "sound plural" has sometimes preserved its bi-literal form: cf. Arabic banima ("sons"), formed from the bi-radical basis ban, as against the tri-radical broken plural 'abna (cf. Blau, 1965, p. 278). 19

5.1.2. Another important morphological innovation shared by the South Semitic languages is the development of the verbal theme qatala. qatala, it seems, is a general West Semitic innovation, yet its development in quite a similar manner in Arabic and Ethiopic (see Fleisch, 1944, especially his general conclusion on pp. 417ff) has, in all probability, to be interpreted as an additional proof for the inclusion of Arabic in the South Semitic group.

5.1.3 I consider the broken plurals and qiitala as the main morphological proof for the affinity of Arabic and South Arabic (including Ethiopic). Other real or alleged shared features are less cogent. Nothing can be inferred from the existence of f, instead of northern p, because the shift p >f, without necessarily applying to other stops, is a universal feature (see Vendryes, 1972, p. 110). In both Arabic and Ge'ez s shifted to s, and.~ to s. This, however, is another case of surprising parallel developments, so often occur­ring in Semitic languages: it has been proven that in Epigraphic South Arabic and modern South Arabic these shifts have not occurred (see the literature cited in Blau, 1970, note 4, and further in Beeston, 1977). The use of qatil for ordinal numbers in Arabic, Ge'ez, modern (and presumably also Epigraphic) South Arabic may be due to shared innovation; it might, however, also be the result of parallel development and contact. Arabic tanwin corresponds to Epigraphic South Arabic tamyim; yet no clear indication for its general use in Ethiopic can be detected. All the South Semitic languages, even those which do not form the causative by *s, form the t-form of the causative by *st. This, however, may be due to phonetic reasons. Whereas no phonetic reasons opposed the replacement of the -~-causative by the hi' -causative, there were phonetic reasons not only for the preservation of the t-form of the s-causative, but even for the replacement of the t-form of the h/'-causative (which sub-

19. Hetzron (1974, p. 183, following Lambert and Greenberg), considers the broken plurals to be a common (Hamito-) Semitic feature, regarding the a-plural of the segolate nouns in Hebrew a remnant of the broken plural as well. This view, however, is not convincing. Forms parallel to these a-plurals are well attested in Arabic, as' aradima from' ard ("land"), darabiit from darb ("blow"), and they must not be interpreted as ancient broken plurals with later addition of the sound plural suffixes, since in Classical Arabic it is the broken plural that expands tc the detriment of the sound plurals. Accordingly, the type guraf ("rooms") has to be regarded as later than gurafiir. Therefore the a-plurals of the monosyllabic ("segolate") nouns have to be consid­ered as original sound plurals, rather than broken plurals with later addition of the plural suffixes.

Page 11: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

HEBREW AND NORTH WEST SEMITIC 31

sisted in Aramaic only; for a different interpretation see Bravmann, 1977, pp. 201-202). For shared lexical features between Arabic and Ethiopic cf., e.g., Littmann (1954, p. 353).

5.1.4 On the other hand, other indications for the alleged special affinity of Arabic with Canaanite and Aramaic, adduced by Hetzron (and following him by Rabin in a lecture delivered at the Seventh World Congress of Jewish Studies, August 1977), are not convincing. Hetzron (1976, pp. 92-93) has rightly stressed what he calls "the principle of archaic heterogeneity" (cf. also, e.g., Blau, 1972a, pp. 90, 120), according to which the relatively most heterogeneous system is considered to be the most archaic, the more homogeneous ones being assumed to have risen as a result of simplification. Yet he does not sufficiently take into consideration that such simplifications may arise independently in various languages as a result of linguistic drift. Hetzron (1974, p. 191; 1975, pp. 93-94) posits two innovative groups: Cen­tral Semitic-including Canaanite, Aramaic, and Arabic; and South Semitic-consisting of South Arabic and Ethiopic; Central Semitic having innovated t-suffixes in the perfect in a homogeneous way, South Semitic k-suffixes, whereas Akkadian has preserved heterogeneous k and t. That the heterogeneous system of Akkadian has been made homogeneous in all the other Semitic languages is quite understandable. Yet, there were only two ways in which this could be achieved: by the preponderance of either t or k. Since the chances in each direction were fifty percent, the chances of inde­pendent development are extraordinarily high, 20 so that it must not be used for genetic grouping.

Hetzron (1974, pp. 189-190; 1976, pp. 94-95) has, quite ingeniously, called attention to the heterogeneity of the prefix-vowels in the prefix tense of G in Akkadian, as against their uniformity in the other Semitic languages. These split into two branches: South Semitic South Arabic and Ethiopic), which completely eliminated the open vowel a from the system; and "Central Semitic,'' which originally used a before thematic i/u, but i before thematic a; later Arabic generalized a and ''the rest submitted the prefix vowels to a reduction process. " 21 Again, it is very difficult to use this feature for genetic classification. It is not even quite certain that the heterogeneity of Akkadian is

20. Through the influence of the corresponding pronominal suffixes, new suffixes penetrated into the second person of the perfect in various dialects (see Noldeke, 1904b, pp. 21-22), an additional indication for the wide range of independent parallel development.

21. So Hetzron ( 1976, p. 95). This, to be sure, is not entirely accurate in Hebrew, but this fact does not affect the problem treated.

Page 12: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

32 JOSHUA BLAU

archaic. Hetzron himself (1976, p. 93) cautioned against using heterogeneity as an indication of an archaic feature if one can find a clear conditioning for differentiation. This, indeed, can be detected in this case: the prefix ni might have been originally na, which later became ni through the impact of the corresponding separate personal pronoun *nihnu (see Ungnad, cited by Hetz­ron, 1974, p. 190, note 10); whereas i might have risen from ya through assimilation(forya >iinO!dAkkadian, see Gelb, 1961,p.122). Yet even if the Akkadian heterogeneity turns out to be archaic, nothing may, in my opinion, be inferred from it for the genetic classification of West Semitic. First, it may well be that the opposition yiqtal:yaqtilul is a general West Semitic phenomenon, characteristic not only of "Central" Semitic, but of South Semitic as well. As is well known, this opposition disappeared also in "Central" Semitic, leaving a few vestiges in Hebrew and only slight traces in Aramaic and Arabic. Accordingly, its absence in most branches of South Semitic (Ethiopic, modem South Arabic) may well be due to later develop­ment, just as it totally disappeared in modem Arabic dialects, whereas we do not know anything of the prefix vowels in Epigraphic South Arabic, the only other alleged branch of South Semitic. On the other hand, this opposition might have emerged independently in different dialects. It is not unreasonable to assume that in some dialects the prefix vowel a prevailed, to become i, through dissimilation, when preceding a, 22 thus giving rise to yiqtal:yaqtul. Yet the development might have been much more involved; the possible intricacy of this process may easily be learned from Grotzfeld's attempts (1964, pp. 28-31) to explain the shift of original a to i (and later, in certain positions, its reduction) in the dialect of Damascus both medially and in affixes. At any rate, one should rather refrain from using the differences in the prefix-vowels of the prefix-tense Gas an indication for genetic classification.

Even less convincing is Hetzron's suggestion (1976, p. 103) to group Arabic and Hebrew together within his alleged Central Semitic, on the grounds that these two languages are "innovative" as against the more ar­chaic Aramaic, insofar as the feminine plural endings in the second and third persons of the verbal conjugations are concerned. The ending is -ii in Akka­dian and Ethiopic, -iin (with an added -n) in Aramaic, whereas Arabic and Hebrew exhibit -na!nii. There is, in my mind, no doubt whatsoever that Arabic and Hebrew reflect independent parallel development. Hetzron him­self saw the reasons for the Hebrew and Arabic innovation quite clearly (in note 25): it was necessitated by the homonymy with the dual ending, and was based on the ending of the corresponding independent pronouns. He even

22. For such a dissimilation in later biblical Hebrew, see Blake (1950).

Page 13: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

HEBREW AND NORTH WEST SEMITIC 33

adduced parallel cases from modem South Arabic, which he rightly consid­ered to be independent innovations. Exactly in the same way, the Hebrew and Arabic innovations were also independent.

5.1.5 So far we have dealt with the features adduced by Hetzron for the alleged genetic grouping of Arabic with Canaanite and Aramaic. In the fol­lowing section we shall adduce some features, not hitherto proposed-shared by Arabic, Canaanite, and Aramaic-which cannot, in my opinion, be used for genetic classification either.

An outstanding feature is, prima facie, the shift of the feminine suffix -at to -ah/-a/-ii, attested in North West Semitic and Arabic. Yet a closer look reveals that this surprising conformity is by no means complete, and because of the different dates at which these suffixes occurred, has to be interpreted as due to parallel development. This shift is totally absent from Ugaritic, yet occurs in Hebrew in both nouns and verbs. In Phoenician it is attested in the verb yet absent in the noun. 23 In Aramaic and modern Arabic dialects, on the other hand, it is absent in the verb, 24 yet attested in the noun. In Classical Arabic it occurs in nouns in the pausal forms25 (in the form -ah), but is absent in verbs and in nouns in context forms. The gap in the dates at which this phenomenon occurs in the various languages is even greater. In Classical

23. In Moabite, too, prima facie -at is preserved in nouns; no examples of the third person feminine singular of the suffix tense are attested. It was in late Punic only that t was omitted in nouns as well (see Friedrich, 1951, p. 99). The reason for the preservation of the t in nouns in Phoenician seems to be that this shift operated in Phoenician in a period in which nouns had still kept case vowels. Therefore the t in nouns was not yet in a final position and, accordingly, not affected by the shift, in contradistinction to the vowelless third person feminine singular of the tense suffix (see Friedrich, 1951, p. 19, note 2). Jn el-Amarna, -al in verbs is still preserved (see Friedrich, 1951. p. 19).

24. Birkeland (1940, p. 97) explained the preservation of -at in the verb by differentiation (darabat > *darabah would have been taken for <darabahu). This explanation, however, is not convincing, since similar forces operated in the realm of the noun as well (*malikat >malikah could be mistaken for malikahii). I would tentatively propose, for both Aramaic and Arabic, that -al shifted to -a(}1), whereas -at (cf. the feminine plural ending in nouns) was preserved. As is well known, the influence of verbs with a final weak consonant on the other verbal classes is quite conspicuous (cf., e.g., Brockelmann, 1908-13, pp. 567, 574), whereas that of nouns with a final weak consonant is much more restricted. It was through the influence of verbs Ill wly, which exhibited-iii in the 3rd person fem. sing., that/ was preserved in verbs. On the other hand, nouns I II w/y terminating in ~-iit shifted to -iih through the influence of nouns with the *-at>ah ending.

25. By no means should one posit pausal orthography outside Arabic. In Arabic itself, it is due to historical chance: since Arabic orthography continued Nabatean Aramaic and the Nabatean forms were mostly identical with the Arabic pausal ones, they were reinterpreted as pausal forms. For particulars see Blau, forthcoming, §4.3.

Page 14: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

34 JOSHUA BLAU

Arabic, as stated, until about 700 A .D., -at was preserved in bound forms, 26

whereas, much more than a millennium earlier, it had already disappeared in the oldest Phoenician and Aramaic inscriptions (in the verb and the noun, respectively). Accordingly, this case may be considered a locus classicus for the extent of proven parallel development.

Hebrew, Ugaritic, Aramaic, and Arabic have, in various degrees, pre­served indications for the forming of the plural of monosyllabic nouns (includ­ing those with a feminine ending) by the insertion of a between the second and third radicals. 27 However, this feature, in all likelihood, is not only an ar­chaism, but it is attested in Ge'ez as well (though only after liquids, where it may also be anaptyctic; see Noldeke, l904a, p. 70).

Nor should one attribute too much importance to the word "what" being expressed by ma/mah in Ugaritic, Canaanite, Aramaic and Arabic (cf. Loewenstamm, 1958-59, p. 74), not only because it may exhibit an archaic feature, but also because a lexical feature such as this may easily be borrowed from dialect to dialect (cf. Singer's summary, 1958, pp. 257-258, as to the interrogative pronouns in the Arabic dialects).

5.2 Not only does Arabic share innovative features with South Arabic (see Sections 5 .1 . I and 5 .1 .2 above), whereas its features common with Canaanite and Aramaic are due to inheritance or parallel development (see Sections 5. I .3 and 5. l .4 above), but it also lacks features characteristic to North West Semitic. Because of the close contact between the Semitic languages, and the large extent of parallel development, one must not claim that these features cogently prove that the North West Semitic dialects once formed a unity in which these features developed. On the contrary, in some cases these features are shared by other Semitic languages as well, thus making the possibility of parallel development even more likely. On the other hand, synchronically at least, the lack of these features in Arabic differentiates it from North West Semitic, making its closer affinity with Canaanite and Aramaic (even histori­cally) rather unlikely.

In North West Semitic, Proto-Semitic w in initial position is represented by y. I do not claim that this feature could not have developed independently

26. Even later is the shift -at >-ohl-Wi in modern South Arabic, since it occurs only in one dialect, in Socotri (see Johnstone. 1975, p. 20). The final h that exists not only in Ar.ibic, but in this dialect as well, makes it quite likely that in the other Semitic languages t first shifted to h, to become afterwards zero.

27. For Hebrew, Aramaic and Arabic, see Noldeke (1904a, p. 70); for Ugaritic cf., e.g., riS ("head"), plural r'asm. Ginsberg (1970, p. 102) was therefore wrong in restricting this feature to Canaanite (including U garitic) and Aramaic only. Cf. also note 19 above.

Page 15: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

HEBREW AND NORTH WEST SEMITIC 35

as well, especially since in Arabic also some isolated roots exhibit both w and y as the first radical (see Noldeke, 1910, pp. 203-206). Nevertheless, it is quite a strong (admittedly phonological) piece of evidence for the unity of North West Semitic, since it is not easy to explain it phonetically (see Nol­deke, 1910, p. 202).

On the other hand, the total assimilation of vowelless n to the following consonant, characteristic of North West Semitic, might have easily emerged independently. It is a well-established feature of Akkadian and is sporadically attested in South Semitic as well. In Classical Arabic it is well attested in sandhi; and for Epigraphic South Arabic cf. Beeston (1962, pp. 16-17). Less convincing is the absence of lqh ("take"), imperfect yqh in Arabic, as against its attestation in North West Semitic, in Ugaritic, Hebrew, Moabite and Aramaic (see Degen, 1969, p. 79 and Hoftijzer, 1965, s.v.).

5.3 It has been claimed several times that South Arabic and Ethiopic (exclud­ing Arabic) exhibit special ties with East Semitic (see the literature cited by Cantineau, 1932, p. 178; and Hetzron, 1974, pp. 183-184). This claim has been refuted by Cantineau ( 1932) and Leslau (1959). It will suffice to refer the reader to these papers, though I do not agree with every item adduced. Thus (in spite of Soden, 1957), I would posit (for "twenty") Proto-Semitic *' ifra, terminating in an original dual ending (=''twice ten''), and for the other tens the plural suffix -un(a)/-in(a). In Akkadian and Ethiopic (in the latter it is the only remnant of the nominative dual ending, as against -e in other cases), as well as in modem South Arabic, if one is entitled to infer such from remnants in Socotri (as cited by Johnstone, 1975, p. 24), ..Q analogically spread over the whole paradigm, whereas in the other Semitic languages it was the plural ending that prevailed. Verbal themes occurring in both Akkadian and Ethiopic formed by a combination of morphemes are, in all likelihood, due to late parallel development; cf., e.g., Middle Hebrew nitqattel as well as similar formations in Arabic dialects (see Brockelmann, 1908-1913, pp. 540-541; Littmann, 1954, p. 359). On the other hand (pace Leslau, 1959, p. 252), the semantic independence of the Ethiopic forms is, in all likelihood, a late development. As to the causative s, as well as 'st (''one''), these forms are, as we know now, attested in U garitic as well. For -ku marking the first person singular of the suffix tense/stative in Akkadian and Ethiopic/South Arabic, cf. Hetzron (1974, p. 191 and 1976, pp. 93-94); for yaqattal etc., see Section 4.2 above; for Proto-Semitic sl9, see Blau (1972b, p. 80).

5.4 Accordingly, one may retain the "accepted" grouping of South Arabic, Ethiopic, and Arabic as South (West) Semitic. Within South (West) Semitic,

Page 16: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

36 JOSHUA BLAU

Ethiopic and South Arabic form a unit, as opposed to Arabic (see, e.g., Cantineau, 1932, pp. 178-185; Hetzron, 1977, p. 12; cf. also the glottalized pronunciation of emphatics in modern South Arabic and Ethiopic [Johnstone, 1975, pp. 6-7]), dubbed by Leslau South East Semitic.

6. Having arrived at the conclusion that Arabic does not form a unit with North West Semitic, we may now proceed to subdivide the North West Semitic languages. We shall not deal with so-called "Amorite," not only because our knowledge is too restricted for any linguistic classification, but also because of our knowledge of the language of Ebia at this time, and it would not be prudent to rush to any conclusions. Taking the accepted division into Canaanite and Aramaic for granted (without broaching the question whether these linguistic groups arose in accordance with the family-tree theory or the wave hypothesis, although we tend to accept the latter; see Sections 2 and 3 above), we shall proceed to the linguistic position of Ugari­tic.

6.1 It is now (see Gordon, 1965, p. 144, who cites additional literature) more or less generally accepted that Ugaritic is a North West Semitic language, yet disagreement exists as to whether to consider it a separate subdivision of North West Semitic or rather to group it within Canaanite. Recently, in a brilliant essay, Ginsberg (1970, especially pp. 102 and 104-106) grouped Ugaritic within Canaanite together with Phoenician, as opposed to Hebraic (consisting of Hebrew and Moabite). I think that Ginsberg has proven his point that Ugaritic is more closely related to Canaanite than to Aramaic. On the other hand, in my opinion, Phoenician and Hebraic belong together ("Canaanite"), as opposed to Ugaritic. 28

6.2 From the features distinguishing Canaanite and Ugaritic from Aramaic adduced by Ginsberg (1970, pp. 103-104), I have not found any that cannot be reasonably attributed either to archaism or parallel development and/or

28. One could dub Ugaritic and Canaanite as "Canaanaic." Yet not only is this a mere play of words, but it also implies genetic affinity in accordance with the family-tree theory (cf. Section 6.2 below). Contrary to Ginsberg (1970, p. 105), I would not simply group the Canaanisms and Canaanite glosses in the Akkadian letters from el-Amarna with Phoenician. It stands to reason that, according to their places of provenance, some were closer to Phoenician, and others closer to Hebrew. Thus, in my opinion, it is not due to mere chance that it is in a letter from Palestinian Gezer that baniti ("I built," in accordance with Hebrew), is attested (el-Amarna 292, 29) as against Phoenician -i:ti (cf. Friedrich, 1951, p. 77). The Canaanite material in these letters, however, is much too restricted to enable us to make these subdivisions.

Page 17: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

HEBREW AND NORTH WEST SEMITIC 37

contact. Nevertheless, one will readily admit that synchronically, Ugaritic exhibits special affinity to Canaanite, though presumably it did not form a genetic group with it within North West Semitic; but rather, in accordance with the wave hypothesis, it more or less developed in the same direction through contact and/or parallel development.

In Canaanite and Ugaritic a suppletive paradigmhlk-ylk ("to go") exists. Yet not only does Akkadian alakum exhibit the apophony a - i, as in verbs I w (see Soden, 1952, p. 128) and Aramaic exhibits suppletive hlklhwk, but the spread of suppletive hlklylk < wlk may be due to mutual contact and/or parallel development. Such parallel development might have started from the imperative *hlik >lik, on the one hand, or from the causative *hahlak >IWlak in the dialects with h-causative (i.e., with the exclusion of Ugaritic), on the other.

The lexical parallels between Ugaritic and Canaanite adduced by Ginsberg are rather impressive. Yet, partly at least, they may well be due to the cultural relation of Ugaritic poetry with Canaanite (Hebrew) poetry.

Nothing must be inferred from the t prefix of the third person feminine plural of the prefix tense. Such an analogical formation, due to the impact of the singular, might well have emerged independently, as it emerged in ancient Arabic dialects (see, e.g., Brockelmann, 1908-13, I, p. 568), in Epigraphic South Arabic (see Beeston, 1962, p. 23) and modem South Arabic (see Johnstone, 1975, pp. 16-17).

Quite a conspicuous isogloss connecting Hebrew with Ugaritic is the merger (synchronically at least) of the D of 11 w/y and the pole! forms etc. The extent of this phenomenon surely gives the impression of later innovation. Parallel development, however, must not be excluded. Several Semitic lan­guages exhibit aversion to doubling w/y (i.e., pawwel, payyel), resorting instead to the doubling of the third radical (for Arabic baynima, daym1ima, Akkadian tukinnii, cf. Blau, 1971, p. 147, note 63; for Aramaic 'etbawrar etc., cf. ibid., p. 148). With this background one has to take into considera­tion also the possibility of convergent development.

In Canaanite (for Phoenician, cf. Friedrich, 1951, pp. 61, 72) certain active participle forms are simply the stem of the perfect (!) inflected as a noun. Such features, however, are not completely absent from other Semitic languages either. In Arabic, adjectives (not real participles) such as /]af ("afraid") from roots ll wly are attested (see Noldeke, 1910, pp. 210-216). Similarly, qatil serves as verbal adjective from intransitive verbs of the qatila pattern (see Wright, 1896-98, I, p. 132) and is, as a matter of fact, much more frequent than the "real" participle qatil. For Aramaic, cf . . ~iii:! ("elder"). Accordingly, Noldeke's cautious remark (1910, p. 209) concern-

Page 18: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

38 JOSHUA BLAU

ing roots II w/y, that this formation "for all that" be considered Proto-Semitic is quite convincing and thus one may regard this feature as an archaism that has been preserved (and perhaps expanded, to some extent).

For the use of the prefix vowel a preceding the thematic vowel ilu of the prefix tense, of the prefix vowel i preceding a, see Section 5. l .3 above. 29

6.3 The features adduced by Ginsberg (l 970, pp. 105-106) as distinguishing Phoenician, el-Amama, and Ugaritic from Hebrew and Moabite are not con­vincing. Lexical features, especially those that are attested in Hebrew as well, though rare and/or poetic, can scarcely be used as decisive proof for genetic classification, the more so when the Ugaritic word pa 'n ("foot") differs from the Phoenician and Hebrew pa'm (rare in Hebrew). The possibility not only of contact, but even of parallel development is considerable: cf., e.g., for "to be," Phoenician, Ugaritic, and Arabic kwn, Hebrew hyy/hwy, and Aramaic hwy, as cited by Ginsberg himself. ytn ("to give") in Ugaritic and Phoeni­cian, is, it seems, due to later independent development, starting from the imperative tin, which, because of its identity with verbs I w, was transferred into that verbal class. Nor is the Hebrew use of perfect and imperfect consecu­tive suitable for genetic classification. It is due to the special Hebrew (and Moabite) development of elements to be found in Phoenician as well, as so ably demonstrated by Ginsberg; cf. also for the consecutive imperfect the 'hrm inscription line 2 wygl, continuing the perfect 'ly. This development surely distinguishes Hebrew and Moabite from Phoenician, being due to a later, presumably shared, innovation; it does not, however, group Ugaritic with Phoenician. A quite conspicuous feature connecting Ugaritic with el­Amarna is the use of tqtlu for the third person masculine plural. This feature, however, is absent not only from Hebrew, but presumably from Phoenician as well (see Friedrich, 1951, p. 54, note I).

6.4 On the other hand, important isoglosses distinguish Ugaritic from He­brew and Phoenician, thus separating these languages (as the Canaanite group) from Ugaritic. Not all the features, to be sure, are equally apt for genetic classification; together, however, in my opinion, they sufficiently warrant a Canaanite group not including Ugaritic.

In my opinion, Ugaritic clearly exhibits the shift l5 >d. The attempt to interpret the Ugaritic d as polyphonic, marking both d and l5 is, to my mind

29. Ginsberg (1970, p. 104) also adduces causative forms like hehbi ("he hid"), occurring in Hebrew, Phoenician and el-Amama. In my opinion, this form has to be considered an archaism, its corresponding form Juprus being attested in Akkadian also; cf. Blau (1971, pp. 152-158).

Page 19: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

HEBREW AND NORTH WEST SEMITIC 39

fallacious (for particulars, see Blau, 1968). The assumption of polyphony does not fit the character of the Ugaritic script, which exhibits added symbols at its end. This sound shift clearly separates U garitic from all the Canaanite dialects known.

It seems that a dialect of U garitic also exhibited the shift ~ >.t. Dietrich et al. (1975) have cited seven words (in ten occurrences) exhibiting the spelling with ~ fort. Five of these occur in one group of texts. In the dialect underly­ing these texts, ~ had shifted to _t; accordingly, there was no difference in whether the sound t was marked by the letter t or ~. In this group of texts it was, from the point of view of the history of the alphabet, the letter ~ that prevailed, just as in Ge'ez the letter z continued South Arabic~, rather than z, ands in the Canaanite alphabet occupies the place of 8, as demonstrated by the Ugaritic alphabet. Two words occur in the archaizing text (Gordon, 1965, text 77) in which ~ also had shifted tot. In this text the use of the letter~ is due to some sort of pseudo-archaism. At any rate, it seems that in at least one Ugaritic dialect, ~, contrary to Canaanite, had shifted tot (whereas the weak sound shift ~ >g exhibits a different dialect). These sound shifts, in my opinion, clearly distinguish Ugaritic from Canaanite.

Other important features common to Hebrew and Phoenician, yet absent from Ugaritic, are: the definite article ha-, the shift ii >o, 30 and the relative pronoun 'a.Serl.Se/' s (for its absence in Ugaritic, see Rainey, 1965-66, pp. 261-263). The first two features are, in my opinion, quite important, whereas I would not attach too much importance to the relative pronoun. I also men­tion here, somewhat with hesitation (since Aramaic and Arabic parallels are not lacking31 ), yt/' ti' et/' ot denoting the definite direct object in Canaanite dialects and lacking in Ugaritic (so far, at least).

7. We have tried to show that the wave hypothesis is much more convincing for the understanding of the classification of the Semitic languages than the family-tree theory (Section 2) and that parallel development characterizes the Semitic languages even more than the Inda-European ones (Section 2.1 ). This may also be inferred from the modem Arabic dialects, the proto-language of which can relatively easily be reconstructed with the help of Classical Arabic (Sections 3, 3.1, 3.2): they attest even to the transfer of unmotivated mor-

30. Since "!" is transcribed in Ugaritic a-na-ku in the quadrilingual word list, it stands to reason that, contrary to Canaanite (and perhaps even Aramaic, see Lipinski, 1976, p. 233a/b), the first person singular of the suffix tense terminated in -tu, rather than in -ti!-ti, as in the Canaanite dialects.

31. See Dion (1974, pp. 164-165).

Page 20: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

40 JOSHUA BLAU

phological innovations (Section 3.2). The basic division of the Semitic lan­guages into East and West Semitic is well established (Section 4). It stands to reason that yaqtulu was a part also of the verbal system of Proto-South East Semitic (Section 5), and perhaps even of that of Akkadian, and thus Proto­Semitic (Sections 4.1, 4.2). Arabic has to be grouped with South Arabic and Ethiopic as "South (West) Semitic" (Section 5. l), because of shared innova­tions between them (Sections 5.1.l, 5.1.2, 5.1.3), whereas the isoglosses connecting Arabic with North West Semitic are not relevant for genetic clas­sification (Sections 5. l.4, 5 .1.5). Moreover, Arabic lacks features charac­teristic of North West Semitic (Section 5 .2). Since South Arabic and Ethiopic do not evince special ties connecting them with Akkadian (Section 5.3), one should retain the "accepted" grouping of Arabic, South Arabic, and Ethiopic as "South (West) Semitic," within which the last two groups form a unit ("South East Semitic"; Section 5.4). West Semitic consists of South (West) Semitic and North West Semitic, the last being sub-divided (disregarding "Amorite") into Ugaritic, Canaanite (including, inter alia, Hebrew and Phoenician), and Aramaic (Section 6. l). Ugaritic is closer to Canaanite than to Aramaic; it cannot, however, be regarded as a Canaanite dialect proper (Section 6). Though the isoglosses marking off Canaanite and Ugaritic from Aramaic do not prove the genetic unity of Canaanite and Ugaritic, neverthe­less they demonstrate, at least synchronically, quite a conspicuous affinity (Section 6.2). The isoglosses separating Ugaritic from Hebrew and Phoeni­cian (including el-Amama) (Section 6.4) are more relevant than those distin­guishing Phoenician (including el-Amama) and Ugaritic from Hebrew (and Moabite) (Section 6.3).

BIBLIOGRAPHY

Barthelemy, A. 1935-69. Dictionnaire arabe-franc,ais. Paris. Beeston, A. F. L. 1962. A Descriptive Grammar of Epigraphic South Ara­

bian. London. -----. 1976. "Notes on Old South Arabian Lexicography X." Le Museon

89:4. 1977. ''On the Correspondence of Hebrew s to Epigraphic South

Arabian s2 ." Journal of Semitic Studies 22:50-57. Bergstrasser, G. 1918-29. Heb raise he Grammatik. Leipzig.

Page 21: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

HEBREW AND NORTH WEST SEMITIC 41

1928. Einfuhrung in die semitischen Sprachen. Miinchen. Birkeland, H. 1940. Altarabische Pausalformen. Oslo. Blake, F. R. 1950. "The Apparent Interchange between a and i in Hebrew."

Journal of Near Eastern Studies 9:76-83. Blanc, H. 1964. Communal Dialects in Baghdad. Cambridge. Blau, J. 1960. Syntax des paliistinensischen Bauerndialekts von Bir-Zet.

W alldorf-Hessen. ----- 1961-62. Review of Kuryl'owicz (1961 ). fa.fonenu 26:68-74. ---. 1965a. The Emergence and Linguistic Background of Judaeo-

Arabic. Oxford. ---. 1965b. Review of A. E. Murtonen, Broken Plurals. L;}fonenu 29:

275-278. ------. 1968. "On Problems of Polyphony and Archaism in Ugaritic Spell­

ing.'' Journal of the American Oriental Society 88:523-526. ---. 1970. On Pseudo-Corrections in Some Semitic Languages.

Jerusalem. ---. 1971. "Studies in Hebrew Verb Formation." Hebrew Union Col­

lege Annual 42:133-158. -----. I 972a. Torat hahege v;}haccurot. Tel Aviv.

l 972b. "Marginalia Semitica II." Israel Oriental Studies 2:57-82. forthcoming. "The Beginnings of the Arabic Diglossia." Afroasiatic

Linguistics. Bloomfield, L. 1933. Language. New York. Bravmann, M. M. 1977. Studies in Semitic Philology. Leiden. Brockelmann, C. 1908-13. Grundriss der vergleichenden Grammatik der

semitischen Sprachen. Berlin. Cantineau, J. 1932. "Accadien et sudarabique." Bulletin de la Socihe de

Linguistique de Paris 33:175-204. Cohen, M. 1924. Le Systeme verbal semitique et!' expression du temps. Paris. Degen, R. 1969. Altaramiiische Grammatik. Wiesbaden. Dietrich, M., 0. Loretz, and J. Sanmartin. 1975. "Untersuchungen zur

Schrift- und Lautlehre des Ugaritischen (III)." Ugarit-F orschungen 7:103-108.

Dion, P.-E. 1974. La Langue de Ya'udi. Otava. Dozy, R. P. 1881. Supplement aux dictionnaires arabes. Leyde. Feghali, M. T. 1919. Le par/er de Kjar 'ahlda (Liban-Syrie). Paris. Fenton, T. L. 1970. "The Absence of a Verbal Formation *yaqattal from

Ugaritic and North West Semitic." Journal of Semitic Studies 15: 31-41.

Ferguson, C. A. 1959. "The Arabickoine." Language 35:616-630.

Page 22: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

42 JOSHUA BLAU

Fischer, W. 1959. Die demonstrativen Bildungen der neuarabischen Dialekte. 's-Gravenhage.

Fldsch, H. 1944. Les verbes a alongement vocalique interne en semitique. Paris.

Friedrich, J. 1951. Phonizisch-punische Grammatik. Rome. Gelb, I. J. 1961. Old Akkadian Writing and Grammar. Chicago. Ginsberg, H. L. 1970. "The Northwest Semitic Languages." The World

History of the Jewish People, II, Patriarchs, pp. 102-124. Tel Aviv. Gordon, C. H. 1965. Ugaritic Textbook. Rome. Grotzfeld, H. 1964. Laut- und Formenlehre des Damaszenisch-Arabischen.

Wiesbaden. Hall, R. A. 1972. "The Reconstruction of Proto-Romance." A Reader in

Historical and Comparative Linguistics, pp. 25-48. Ed. A. Keiter. New York.

Hetzron, R. 1974. "La division des langues semitiques." Actes du Premier Congres International de Linguistique Semitique et Chamito- Semitique, pp. 182-194. Eds. D. Cohen and A. Caquot. The Hague.

---. 1975. "Genetic Classification and Ethiopian Semitic." Hamito­Semitica, pp. 103-127. Eds. J. and T. Bynon. The Hague.

---. 1976. "Two Principles of Genetic Reconstruction." Lingua 38: 89-108.

---. 1977. The Gunniin-Gurage languages. Napoli. Hofner, M. 1943. Altsudarabische Grammatik. Leipzig. Hoftijzer, J. and C. F. Jean. 1965. Dictionnaire des inscriptions semitiques

de l' ouest. Leiden. Jamme, A. 1962. Sabaean Inscriptions from Mahram Bilqis ( Marib). Balti­

more. Johnstone, T. M. 1975. "The Modem South Arabian Languages." Afroasia­

tic linguistics l, 5:1-29. Kaufman, S. A. 1974. The Akkadian Influences on Aramaic. Chicago. Kofler, H. 1940-42. "Reste altarabischer Dialekte." Wiener Zeitschrift zur

Kunde des Morgenlandes 47 (1940):61 130, 233-262; 48 (1941):52-88, 247-274; 49 (1942):14-30, 234-256.

Kuiylowicz, J. 1961. l' apophonie en semitique. Wroclaw. Kutscher, E. Y. 1971. ''Aramaic.'' Current Trends in linguistics 6, Linguis­

tics in South West Asia and North Africa, pp. 347-412. The Hague. Leslau, W. 1959. "The Position of Ethiopic in Semitic: Akkadian and

Ethiopic." Akten des 24. internationalen Orientalisten-Kongresses, Munchen 28 August-4 September. Wiesbaden.

Page 23: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

HEBREW AND NORTH WEST SEMITIC 43

Lipinski, E. 1976. Review of Dion (1974). Bibliotheca Orienta/is 33: 231-234.

Littmann, E. 1954. "Die athiopische Sprache." Handbuch der Orientalistik, Ill, Semitistik, pp. 350-375. Leiden.

Loewenstamm, S. E. 1958-59. "L;:ltorat hakkinnuyim b;'.l'ugaritit );'.)'or hakkana'anit." l;'.lsonenu 23:72-84.

Meyer, R. 1966-72. Hebriiische Grammatik. Berlin. Meillet, A. 1958. linguistique historique et linguistique generale I. Paris. Moran, W. J. 1965. "The Hebrew Language and its Northwest Semitic

Background." The Bible and the Ancient Near East, pp. 59-84. Ed. G. E. Wright. New York.

Noldeke, T. 1904a. "Zur semitischen Pluralendung." Zeitschr(ft far As­syriologie und verwandte Gebiete 18:68-72.

l 904b. Beitriige zur semitischen Sprachwissenschaft. Strassburg. ---. 1910. Neue Beitriige zur semitischen Sprachwissenschaft.

Strassburg. Polotsky, H.J. 1964. "Semitics." The World History of the Jewish People.

I, I, The Dawn of Civilization, pp. 99-111, 357-358. Tel Aviv. Porzig, W. 1954. Die Gliederung des indogermanischen Sprachgebiets.

Heidelberg. Rainey, A. F. 1965-66. "Kelim hadasim 1;'.lheqer ha'ugaritit." La5onenu 30:

250-272. Rabin, C. 1963. "The Origin of the Subdivision of Semitic." Hebrew and

Semitic Studies presented to Godfrey R. Driver, pp. 104-115. Oxford. Rimalt, E. S. 1932. "Wechselbeziehangen zwischen dem Aramaischen und

Neubabylonischen." Wiener Zeitschrift zur Kunde des Morgenlandes 39: 100-122.

Sarauw, C. 1912. "Das altsemitische Tempussystem." Festschrift fur Vi/­helm Thomsen, pp. 59-69. Leipzig.

Singer, H. -R. 1958. Neuarabische Frageworter. Inaugural-Dissertation Er­langen. Erlangen.

Soden, W. von. 1952. Grundriss der akkadischen Grammatik. Rome. ---. 1957. "Die Zahlen 20-90 im Semitischen und der status absolutus."

Wiener Zeitschrift zur Kunde des Morgenlandes 57:24-28. ---. 1959. "Tempus und Modus im Semitischen." Akten des 24. inter­

nationalen Orientalisten-Kongresses, Miinchen. August 28-September 4. Wiesbaden.

Southworth, F. C. 1964. "Family-Tree Diagrams." language 40:557-567. Ungnad, A. and L. Matous. 1964. Grammatik des Akkadischen. Miinchen.

Page 24: HEBREW AND NORTH WEST SEMITIC: … AND NORTH WEST SEMITIC: REFLECTIONS ON THE CLASSIFICATION OF THE SEMITIC LANGUAGES By JOSHUA BLAU Hebrew University of Jerusalem 1

44 JOSHUA BLAU

Vendryes, J. 1972. "Some Thoughts on Sound Laws." A Reader in Histori­cal and Comparative Linguistics, pp. 109-120. Ed. A. Keiter. New York.

Weinreich, U. 1953. Languages in Contact. New York. Wright, W. 1896-98. A Grammar of the Arabic Language3 . Cambridge. Zimmem, H. 1917. Akkadische Fremdworter als Beweis jiir babylonischen

Kultureinfluss 2 • Leipzig.